AMYLOID FIBRIL FORMATION IN ALZHEIMER's DISEASE
by
Joseph Timothy Jarrett
B. S. Chemistry, University of Michigan (1988)
Submitted to the Department of Chemistry in Partial Fulfillment of the
Requirements for the Degree of DOCTOR OF PHILOSOPHY
at the
Massachusetts Institute of Technology May 1993
@ 1993 Massachusetts Institute of Technology All rights reserved
Signature of Author Certified by I, Department of Chemistry May 19, 1993 Peter T. Lansbury, Jr. Associate Professor of Chemistry Thesis Supervisor
Accepted by.
~,
(
Glenn A. BerchtoldChairman, Departmental Committee on Graduate Students
AfCH!VES MASSACHUSETTS INSTITUTE OF TFr'NMOI MGy
[JUN
29 1993
x innAOICC ,,2
This doctoral thesis has been examined by a Committee of the Department of Chemistry as follows:
Professor Joanne Stubbe
Chairperson
Professor Peter T. Lansbury, Jr.
7-V
7/i'
Thesis Supervisor Professor Jonathon King/
Department of/iologyProfessor Jun Liu
151
References
(1) Spencer, R.; Halverson, K.; Auger, M.; McDermott, A.; Griffin, R.; Lansbury, P., Jr. Biochemistry 1991, 30, 10382.
(2) Eanes, E.; Glenner, G.
J.
Histochem. Cytochem. 1968, 16, 673.(3) Fraser, R.; MacRae, T. in Conformation in Fibrous Proteins and Related
Synthetic Polypeptides; Academic Press, New York, 1973.
(4) Termine,
J.;
Eanes, E.; Ein, D.; Glenner, G. Biopolymers 1972, 11, 1103. (5) Kirschner, D.; Abraham, C.; Selkoe, D. Proc. Natl. Acad. Sci. USA 1986, 83,503.
(6) Fraser, P.; Nguyen,
J.;
Surewicz, W.; Kirschner, D. Biophys.1.
1991, 60, 1190. (7) Kirschner, D.; Inouye, H.; Duffy, L.; Sinclair, A.; Lind, M.; Selkoe, D. Proc.Natl. Acad. Sci. USA 1987, 84, 6953.
(8) Fraser, P.; Nguyen,
J.;
Inouye, H.; Surewicz, W.; Selkoe, D.; Podlisny, M.; Kirschner, D. Biochemistry 1992, 31, 10716.(9) Marsh, R.; Corey, R.; Pauling, L. Biochim. Biophys. Acta 1955, 16, 1. (10) Arnott, S.; Dover, S.; Elliot, A.
J.
Mol. Biol. 1967, 30, 201.(11) Lansbury, P., Jr. Biochemistry 1992, 32, 6867.
(12) Halverson, K., The Molecular Determinants of Amyloid Deposition in
Alzheimer's Disease, Ph.D. Thesis, Massachusetts Institute of Technology,
1992.
(13) Halverson, K.; Fraser, P.; Kirschner, D.; Lansbury, P., Jr. Biochemistry 1990, 29, 2639.
(14) Halverson, K.; Sucholeiki, I.; Ashburn, T.; Lansbury, P., Jr.
J.
Am. Chem.Soc. 1991, 113, 6701.
(15) Krimm, S.; Bandekar, J. Adv. Protein Chem. 1986, 38, 181.
(16) Wilder, C.; Friedrich, A.; Potts, R.; Daumy, G.; Francoeur, M. Biochemistry 1992, 31, 27.
(17) Eberhardt, E.; Loh, S.; Hinck, A.; Raines, R.
J.
Am. Chem. Soc 1992, 114,5437.
(18) Richardson,
J.; Richardson, D. in Prediction of Protein Structure and the
Principles of Protein Conformation; Fasman, G. D.; Plenum Press, New York,
1989; p 1.
(19) Ramachandran, G.; Mitra, A.
J.
Mol. Biol. 1976, 107, 85.(20) Radzicka, A.; Pederson, L.; Wolfendon, R. Biochemistry 1988, 27, 4538. (21) Grathwohl, C.; Withrich, K. Biopolymers 1976, 15, 2043.
(22) Drakenborg, T.; Dahlqvist, K.-I.; Forsen, S. 1. Phys. Chem. 1972, 76, 2178. (23) Larive, C.; Rabenstein, D.
J.
Am. Chem. Soc. 1993, 115, 2833.(24) Levitt, M.
J.
Mol. Biol. 1981, 145, 251.(25) Privalov, P. Adv. Prot. Chem. 1979,33, 167.
(26) Pauling, L.; Corey, R. Proc. Natl. Acad. Sci. USA 1952, 38, 86.
(27) Kessler, H.; Anders, U.; Schudok, M.
J.
Am. Chem. Soc. 1990, 112, 5908.(28) Sukumaran, D.; Porok, M.; Lawrence, D.
J.
Am. Chem. Soc. 1991, 113, 706.(29) Dyson, H.; Rance, M.; Houghten, R.; Lerner, R.;Wright, P.
J.
Mol. Biol.3
AMYLOID FIBRIL FORMATION IN ALZHEIMER'S DISEASE by
Joseph Timothy Jarrett
Submitted to the Department of Chemistry on May 19, 1993 in partial fulfillment of the requirements for the Degree of Doctor of Philosophy
ABSTRACT
Amyloid diseases are characterized by the presence of highly insoluble protein deposits in specific tissues or throughout the body. The amyloid protein in Alzheimer's disease is the
1
protein, a ca. 4 kDa protein which derives from a 110-135 kDa precursor protein. TheP
protein is produced as an excreted protein in cell culture, and has been detected in the CSF and blood of healthy individuals. AD amyloid contains several variants of this protein, differing at their N- and C-termini. The variants differing at the C-terminus have been reported to have different properties:p1-39
andp1-40
are more soluble than 11-42 and P1-43. We have studied peptides derived from a bacterial protein which are similar to the C-terminus of thep
protein, and have shown that amyloid fibril formation is a nucleation-dependent assembly process. The kinetics are characterized by a lag phase, during which the peptide is apparently soluble, followed by a growth phase, during which fibrils are rapidly formed. Seeding with fragmented fibrils results in new fibril formation, bypassing the rate-limiting nucleation step. Fibril formation requires highly specific hydrophobic interactions, as demonstrated by the sequence specificity of seeding. In peptides which model the C-terminus of theP
protein, the longer variants 26-43 and P26-42 form fibrils immediately while the shorter variants 126-40 and 126-39 have lag times of hours to days before fibril formation occurs. Addition of fibrils derived from any of theP
proteins results in seeding of the slower nucleating variants. We have also studied the affect of the C-terminus on the structure. Using 13C labeled peptides and solid-state NMR, we have shown that P26-40 and 126-43 have a cis amide between Gly37 and Gly38. Using isotope-edited FTIR, we show that there is considerable structural variation between 126-40 and $26-43 in the vicinity of this cis amide. $26-43 appears highly ordered while the slower nucleating $26-40 appears disordered. Finally we show that hydrophobic effects in solution may stabilize the cis amide, and this may play a critical role in the mechanism of amyloid formation.Thesis Supervisor: Dr. Peter T. Lansbury, Jr.
Associate Professor of Chemistry
Table of Contents
A cknow ledgm ents ... 7
List of A bbreviations ... 8
Chapter 1 Alzheimer's Disease and the Amyloid
p
Protein... 9Pathological features of AD ... 10
A m yloid Plaques ... 12
The Source of the
p
Protein... 14The
p
Protein is Produced as a Normal Extracellular Protein...18P
Protein Aggregate May Be Neurotoxic... 19R eferen ces... . 22
Chapter 2 Amyloid Fibril Formation is a Nucleation-Dependent Assembly P ro ce ss ... 24
A m yloid osis ... . 25
A m yloid D isease ... 25
Mechanism of In Vivo Amyloid Formation. ... 28
Fibril Formation by Peptides Derived from the E. coli OsmB Protein... 30
Sequence Similarities Exist in Amyloid Proteins... 30
OsmB Derived Peptides Form Amyloid Fibrils... 33
FTIR of Peptide Fibrils... 37
Circular Dichroism of Peptide Solutions ... 37
Solubilities of OsmB Derived Peptides are Similar ... 40
Time-Dependent Aggregation of Peptide Solutions... 42
Seeding of Supersaturated Peptide Solutions ... 44
Su m m ary ... . 47
Protein Self-Assembly Processes... 48
Tubulin Polymerization - Thermodynamically-Limited N ucleation ... . 49
Hemoglobin S Polymerization - Kinetically-Limited Polym erization... 54
General Mechanism for Nucleation-Dependent Amyloid Form ation . ... . . 59
Experim ental... 69
M aterials. ... . . 69
Protein Sequence Search. ... 70
Peptide Synthesis, Purification, and Characterization. ... 70
Electron Microscopy. ... 72
Fourier-Transform Infrared Spectroscopy... 73
Circular Dichroism Spectroscopy. ... 73
Peptide Solubility. ... 75
Kinetic Aggregation Studies... 75
Problems Associated With Kinetic Aggregation Studies ... 76
R eferences... 80
Chapter 3 Kinetic Studies of Amyloid Formation by
1
Protein Derivatives ... 83Solution Properties of the $ Proteins... 84
Aggregation Properties of Proteins ... 86
Aggregation of Truncated C-Terminal $ Protein Variants ... 88
Model Peptides Form Amyloid Fibrils ... 89
Solubility is Slightly Sensitive to C-Terminus ... 91
Time-Dependence of Aggregation is Sensitive to C -T erm inus ... . 93
Seeding Results in Immediate Aggregation... 95
$26-39 and 126-40 are Seeded by All of the C-Terminal V arian ts ... . . 97
Fibril Growth is Reversible ... 97
Fibrils Become More Stable Over Time... 100
Model for Amyloid Formation in AD ... 100
E xp erim en tal... 103
Synthesis and Purification of Peptides... 103
Thermodynamic Solubility ... 106
Kinetic Aggregation Studies ... 106
Seeding of Peptide Aggregation ... 106
Fibril Disaggregation and Fibril Stability ... 107
R eferen ces ... 108
Chapter 4 Structural Studies of Amyloid Formed by
1
Protein Derivatives ... 110Structure of
1
Protein Amyloid ... 111Effect of C-Terminus on Solid-State Structure ... 113
FTIR of Unlabelled Peptides ... 113
FTIR of Peptides Labelled at Glycine 38...117
Gly37-Gly38 Distance Measured by Solid-State NMR ... 120
Structural Model of C-Terminal Effects ... 122
Effect of Seeding on Structure of Fibrils ... 124
Model for C-Terminal Effects on Kinetics and Structure ... 140
E xp erim en tal...142
6 FTIR spectroscopy ... 146 Solid-State N M R Spectroscopy ... 147 Solution N M R Spectroscopy... 148 References...151 A ppendix ... 153 1H -N M R Spectra ... 154 M ass Spectra ... 169
Acknowledgments
Graduate school has been the most difficult and challenging period in my life. Yet when I think about other things I might have done over the last five years, everything else pales in comparison to the experiences I've shared here at MIT. Both research and life here move at a frenetic pace, and it's only now, when I have time to think about what I've been through, that I realize how good life has been.
Peter Lansbury has contributed to my life and career in untold ways. He has given me complete freedom in the lab, allowing me to run with new ideas when the creative urge strikes. Most of the experiments described in this thesis are a result of this freedom. On very rare occasions he has reigned me in long enough to characterize a peptide or repeat an experiment more carefully. And when big purple stains covered my hood, or when a liter of phosgene suddenly appeared in the lab, Peter's casual question would be "So Joe, any interesting results lately?" Peter also inadvertently introduced me to a pretty young undergrad who would later become my wife. I don't know how to thank him.
One of the first people I met after arriving at MIT was Beth Berger. Who could have predicted that our lives would become so intertwined. As labmates we shared everything: benches, hoods, glassware, projects, problems ... Often Beth would remember to stop experiments for me long after I'd gone home. And on top of that, Beth is one of the best friends I've made at MIT. Thanks.
One of my true mentors at MIT has been Kurt Halverson, who has since returned to the land of his forefathers, Minnesota. My entry into amyloid research was the result of a casual suggestion Kurt made over a beer at the Muddy. Kurt taught me valuable lessons that are so hard to learn on your own: how to cast a fly rod, how to catch a trout, how to true a wheel, how to plan a bike trip, and most important of all, the proper way to tell a joke or a story in a bar (I'm still learning).
Jon Come rounded out this trio of insane bikers. My body still hurts from our gut-wrenching rides through the Adirondack Mountains. I think Jon and I hold some kind of a record after stumbling home drunk together 8 days in a row in June of '90. Jon has also been a healthy critic of my wild speculative research interpretations, which has often forced me to think about my results more than I wanted to, but has made for interesting (and long) group meetings.
The rest of the Lansbury Lab has contributed in separate ways to my research and my life. It's never very hard to find someone to talk with about a research problem or to unwind with at the Muddy after a long (or short) day.
My Mom and Dad have always been there when I needed them. They have always encouraged me to be better than I thought I was capable of. However, despite my Mom's efforts, as this thesis has shown, I will never be able to stop procrastinating.
Most graduate students go through a period when they feel like leaving MIT. Just as I was starting to feel this way, I met Beth. Everything in my life changed. When my attitude reaches new lows, Beth's faith in me never wavers. I think that more than anyone else, she is the reason I have finished graduate school. Beth, now that this thesis is finished, perhaps I'll be able to spend some time with you. But I hope our life together never becomes "normal."
List of Abbreviations
Ac acetyl
AD Alzheimer's disease
APP amyloid precursor protein
Boc t-butoxycarbonyl
BOP benzotriazol-1-yl-oxy-tris-(dimethylamino)phosphonium
hexafluorophosphate
CAS Chemical Abstract Service
CSF cerebrospinal fluid
DIEA diisopropylethylamine
DMF dimethylformamide
DMS dimethylsulfide
DMSO dimethylsulfoxide
FABMS fast atom bombardment mass spectrometry
Fmoc fluorenylmethoxycarbonyl
FTIR Fourier-transform infrared spectroscopy
HFIP hexafluoro-2-propanol
HPLC high pressure liquid chromatography
MBHA methylbenzhydrylamine
MS mass spectrometry
NMR nuclear magnetic resonance
PAM phenylacetamidomethyl
PDMS plasma desorption mass spectrometry
PrP prion protein
PyBroP bromo-tris-pyrrolidino-phosphonium hexafluorophosphate RPHPLC reverse phase high pressure liquid chromatography
TDC transition dipole coupling
TFA trifluoroacetic acid
TFE trifluoroethanol
Chapter 1
Alzheimer's Disease and the Amyloid
P
Protein
The human body deteriorates rapidly in old age, resulting in changes in appearance and physical abilities.1 Wrinkled skin, stiff joints, brittle bones, and reduced lung function are all due to changes in connective tissue proteins, in particular cross-linking of collagen. Most other organs deteriorate, resulting in impaired function. In particular, the kidneys lose nephrons, resulting in a decrease in the ability to remove waste products from the blood. The continual presence of waste products results in the overload and weakening of the immune system.1 Yet throughout such profound deterioration of the body, many individuals are capable of maintaining a sharp, active mind. Winston Churchill, for instance, was Prime Minister of Britain from age 65-80, during some of the most politically difficult years of this century. Pablo Picasso maintained his artistic productivity until his death at age 92.1 Other individuals may show signs of senility (from the Latin senex: old man) in their 80's, yet as the word implies, we have long considered a slight deterioration of the mind as a normal part of late stages of the aging process. However, in certain individuals, the aging process seems reversed. Starting as early as their fourth or fifth decade, patients with Alzheimer's disease (AD) begin to lose their cognitive functions, including memory, language use, perception, abstract thought, and the ability to solve problems and make judgments.2 Yet they maintain relatively healthy bodies. In essence, they lose only those abilities which are characteristic of human existence,
the ability to think abstractly and to make judgments based on the memory of past experiences.
The loss of memory is not always the most noticeable symptom of AD. Alzheimer described his first patient: "The first noticeable symptom of illness ... was suspiciousness of her husband .... believing that people were out to murder
her, [she] started to scream loudly. . . .At times she is totally delirious .. ."3,4 While memory and cognitive skills may initially deteriorate, patients often deny or hide these problems from their family; the first signs of AD are often psychotic symptoms such as withdrawal, explosiveness, delusions, and hallucinations,5 which derive from a loss of control over thoughts and actions. It is only later, as cognitive abilities become seriously impaired, that awareness fades and patients require constant care. The occurrence of AD increases with age, doubling every 5 years. Approximately one-third of individuals over the age of 85 have diagnosed AD.2 The prevalence of AD has increased in our society due to improvements in sanitation and medicine which have led to a general aging of our population.1 In 1987, ~10% of our population was over 65,6 while -1% of our population had AD. The estimated costs of caring for these persons was over $40 billion/year.6 This amount will increase dramatically as we approach the 21st century.
Pathological features of AD
Although presenile dementia had been recognized since the mid 19th century, Alzheimer took advantage of advances in cell staining to observe pathological changes in brain sections upon autopsy. Upon silver staining cortical slices, he noted extensive neuron death in the cerebral cortex. In addition, he noted two features which are now considered the hallmarks of AD: intracellular neurofibrillary tangles and extracellular neuritic plaques. Plaques
degenerating nerve terminals and the remnants of dead cells.2 Cerebrovascular amyloid is usually found as sheaths which form around the blood capillaries of the meninges, cortex, and hippocampus, often resulting in damage to the blood-brain barrier. Neurofibrillary tangles are composed primarily of hyperphosphorylated forms of the protein tau, which is a major component of the cytoskeleton. They are also observed in other neurological disorders, and are believed to be a general indication of cell death, rather than a cause of AD.7 Granulovacuolar bodies, a type of inclusion body, are also found in certain hippocampal neurons.2 Finally, there is considerable atrophy of the cerebral cortex which results in a loss of 10-15% of total brain weight relative to age-matched controls.2
The understanding of Alzheimer's disease has been hampered by the inability to diagnose the disease until autopsy. The features of the disease noted by Alzheimer require brain section and staining to observe plaques and tangles. It is now known that AD causes ~55% of cases in a class of disorders known as presenile dementia.2 However, there is some debate over whether AD is a single disease, or a group of disorders which manifest similar symptoms and pathological changes.8 Systematic diagnosis now relies heavily on standard tests of cognitive skills. The first of these was developed by Blessed, who was able to show correlations between dementia and the extent of cortical cell death.9 Those patients who had no history of stroke showed a strong correlation between mental-status scores and the number of neuritic plaques.9 This was the first and is still the strongest evidence regarding the cause of AD symptoms: dementia seems to be caused by extensive cortical cell death, and in AD, this cell death is accompanied by plaque formation.
Amyloid Plaques
Plaques are observed in silver-stained cortical tissue sections as 50-100 pm diameter dark, spherical masses surrounded by dystrophic neurites.7 The tissue surrounding these plaques is characterized by dead neurons, primarily cholinergic neurons, and by the presence of intracellular 10-20 ptm neurofibrillary tangles. The composition of the plaques has remained ambiguous, and is complicated by difficulties in separating plaques from the surrounding tissue without losing weakly bound components. The plaque cores have recently been successfully purified by centrifugation followed by fractionation of the pellet using a fluorescence-activated cell sorting system.10 11 The major inorganic components include aluminum, iron, calcium, and silicon.12-14 The protein components which have been identified include the
P
protein, cc1-antichymotrypsin,1 5 serum amyloid P,16 heparan sulfate proteoglycan,protease nexin-I, and cholinesterases, along with various unidentified trace components.17 The primary organic component is the
P
protein, accounting for >60% of plaque core protein.14The
P
protein was first extracted from vascular plaques by Glenner and Wong using 6 N guanidine HCl.18 They showed that this was a ca. 4 kDa protein and they reported a partial N-terminal sequence. An identical sequence was later identified in neuritic plaque cores,19 and the sequence was extended to 40 residues inp
protein derived from meningovascular amyloid.20 The plaque protein from the neuritic plaques proved more difficult to solubilize or sequence. ThisP
protein is insoluble in 6 N guanidine HC1, but has been solubilized in formic acid. One laboratory was able to sequence the first 24 residues, and confirmed that the plaque core protein was the same sequence as the vascular amyloid protein.19 However, sequencing proved difficult due to the presence ofFigure 1.1
Sequences of
P
proteins found in amyloid deposits.'4FRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVVIA DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVVIA
euritic
FRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVV
DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVV DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVV AEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVVvascula1
DAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGV
AEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVheterogeneity at the terminus, possibly also including chemically modified
N-termini.7,21
Recently, two independent laboratories have characterized the contents of the vascular and neuritic plaques using mass spectrometry and N-terminal chemical sequencing. David Miller and coworkers along with loannis Papayannopoulos of the MIT mass spectrometry facility carefully characterized the components of cerbrovascular amyloid and amyloid plaque core proteins.'4 They found that the neuritic plaques contained 25% protein by weight. Of this protein, 90% was formic acid soluble, and 70% (by wt., -95% by mole fraction) of the acid-soluble fraction was
P
protein. The neuritic plaques contained a heterogeneous mixture ofP
proteins. (Figure 1.1) The N-terminus wasdetermined by chemical sequencing and was found heterogeneous, with peptides beginning at each of the first 10 residues. The primary component began at Phe-4. After chymotryptic digest, the C-terminal peptides were purified and identified as a mixture of Val-40 and Ala-42; the relative amounts were not ascertained. The cerebrovascular amyloid was less heterogeneous: the N-terminus was a mixture of Asp-1 and Ala-2, while the C-N-terminus was a mixture
Figure 1.2
Diagram of the f Amyloid Precursor Protein.
C C
p
secretase H H cleavage cytoplasm0 0
Kunitz Proteasej
Protein
Inhibitorof Val-39 and Val-40. Another laboratory has reported similar results,22 although their study was not as meticulous with regard to the purification and quantitation of the
P
proteins. In partially purified cerebral amyloid, they found peptides with heterogeneity at the N-terminus, and C-termini which ended at Val-39, Val-40 and Thr-43.The Source of the
P
ProteinAfter Glenner reported a partial sequence for the vascular plaque, it was assumed that the small amyloid protein must derive from a larger precursor, and the race was on to clone and sequence the full length precursor. Several groups identified the sequence for a 110-135 kDa protein termed the
P
amyloid precursor protein (APP).23 27 (Figure 1.2) This protein is expressed in three forms through differential mRNA splicing: a 695, 751, and 770 residue forms differing by two small inserts. The 56 residue insert in the longer APPs corresponds to a known protease inhibitor sequence. All three APPs have a long extracellular section, which is N- and 0-glycosylated and tyrosyl-sulfated in human cell cultures.7,21Figure 1.3
Mutations in APP linked to amyloid disease.
--EVKMDAEFRHDSGYEVHHQKLVFFAEDVGSNKGAIIGLMVGGVVIATVIVIT-
-NL GQ I
'.
double Fmutation G
The
P
protein sequence begins at residue 597 (APP-695) in the extracellular domain and extends into the predicted transmembrane region. The protein also contains a short cytoplasmic tail which contains a signal sequence for endocytotic degradation. The structure or function of the APP proteins are unknown, although it has been suggested that the membrane-bound APP may have a role in cell adhesion.21Identification of the APP protein as the source of the
P
protein in plaques led other researchers to probe whether some cases of AD could be explained by mutations in the APP DNA sequence. The APP gene was found on the long arm of chromosome 21. In Down's syndrome, three copies of chromosome 21 result in an overproduction of APP; amyloid plaques are found in the brains of deceased Down's syndrome patients even in early childhood.1 7 In addition,genetic-linkage studies have shown that in some cases, early-onset AD could be linked to genetic markers on chromosome 21.28 In 1990, it was discovered that a rare disease, hereditary cerebral hemorrhage with amyloidosis, Dutch type, is caused by a point mutation in the APP gene, substituting glutamine for glutamic acid at position 693. This residue is at position 22 of the
P
protein from AD.29 (Figure 1.3) Independently, several families were discovered with mutations in APP which occur within the transmembrane sequence, but just outside of theP
protein sequence. These result in the rather conservative mutation of valine toisoleucine, phenylalanine, or glycine.30 Additional mutations have since been discovered in the APP gene, summarized in Figure 3. Unfortunately, these mutations account for less than 1 percent of all AD cases. Also, there is evidence that other families may have mutations on other chromosomes which also cosegregate with the disease, but are unrelated to the APP gene. However, the genetic evidence supports the theory that the protein is causative in at least some instances of Alzheimer's disease.
The sequence of APP suggests that the
p
protein is liberated by at least two protease cleavages, one N-terminal to Asp-671 in the extracellular domain, and one in the vicinity of Val-709 to Thr-713 located within the transmembrane domain. (Figure 1.4) In certain cell lines, as much as 80% of APPs are degraded nonspecifically via a lysosomal pathway.23 However, -20% of APPs undergo cleavage to release a soluble N-terminal domain (685-745 kDa)31 while the C-terminal domain undergoes endocytosis and is degraded in lysozomes.7 (Figure 1.4, center pathway) The secretase which carries out this cleavage has been shown to have little sequence specificity, but rather cleaves APP ~12-13 residues outside the membrane, and seems to require a specific structure which was disrupted by the presence of a proline mutant.23,32 Cleavage by this proteaseseems to be the dominant processing pathway for APP; this pathway leads to truncated
P
proteins, which may also be amyloidogenic, although they have not been detected in mature plaques.'4 Two other pathways have been discovered in cell culture which are believed to release the intactP
protein. One cleavage event generates C-terminal APP fragments which contain the entireP
protein sequence.33-35 (Figure 1.4, cleavage I) These protein fragments were foundassociated with lysosomes, and have the potential of generating the
P
protein intracellularly. The corresponding soluble APP derivatives which stopped at Met-596 (APP-695) were found in cell medium and CSF.36 Another cleavageFigure 1.4
Processing of APP by the
P
secretase, proteinase I, and proteinase II pathways. Numbers refer to residues in APP-695 for production of $1-42.1/|
695DAPP
11 6384
11 1596 597M 638 613 p3 I 612 1 596 695 597-111[
695 613K 638 Amyloid Plaque 639 695 597w 638event generates the C-terminus of the
P protein in cell culture. (Figure 1.4,
cleavage II) Together with the first cleavage, this generates the intactP
protein.37,38 The N-terminus was determined to be Leu-(-1)38 or Asp-1 and Leu-17.37 The precise C-terminus was not ascertained; the approximate molecular weights were 4 and 3 kDa. Finally, the extracellular domain of APP has been detected which apparently contains the intactP
protein at its C-terminus.39 Thisfinding indicates that the 3 protease events are independent of each other, and 639 695
there is no necessary proteolysis sequence which must be followed. All of the possible cleavage products have been observed in cell culture. The processing pathways are summarized in Figure 1.4.
The P Protein is Produced as a Normal Extracellular Protein
Detection of the P secretase products led to the assumption that this was the "normal" processing pathway for APP.23 In the search for the "abnormal" cleavage products, a surprising result was obtained. Using antibodies to synthetic P1-40, the P protein was detected in both cell culture medium37,38,40 and
in biological fluids from both normal and AD patients.38,40 Selkoe and coworkers37 transfected human kidney cells with APP-695, and detected 4 kDa and 3 kDa proteins with P protein immunoreactivity. Radiosequencing revealed the N-terminus was Asp-1 for the 4 kDa protein, and Val-18 for the 3 kDa protein. The C-termini were not determined. Younkin and coworkers38 transfected human leukemia cells with an 11.4 kDa C-terminal derivative of APP and with APP-695. Both cell lines produced P protein in the cell medium. Radiosequencing indicated that this P protein began at Asp-1; the C-terminus was not determined. The amount of
P
protein was estimated to be 40% of the 10 kDa fragment derived from the P secretase cleavage. Since the secretase is estimated to account for -20% of APP processing,23 this would indicate that the production of the P protein may account for as much as 8% of APP processing. CSF from 5 of 7 AD patients and 3 of 7 normal age-matched controls also contained detectable amounts off
protein.38 The P protein was also detected inbiological fluids by Seubert and coworkers.36 Using a monoclonal antibody affinity column, they purified P protein from the CSF and blood plasma of normal humans, dogs, guinea-pigs, and rats. They estimated the concentration
identified by mass spectrometry as
p1-40.
However, several other proteins with high affinity were not identified./ Protein Aggregate May Be Neurotoxic
While the presence of amyloid plaques in AD brains is an important hallmark of the disease, any model for the progression of the disease must account for cell death, since the observed symptoms of AD are caused by the death or atrophy of cholinergic neurons in the cerebral cortex.41 Cell death seems to be more pronounced in regions near plaques, but has been noted in tissue that did not contain plaques. Perhaps the most controversial issue in AD research has been the cause of neuronal death. Bruce Yankner found his work thrust under the microscope of his peers when he demonstrated that synthetic
P
protein caused neuronal death both in human neuronal cell culture and in rat brain.42 Other researchers spent a great deal of time and money trying to reproduce Yankner's results, many with little success.43 Unfortunately, researchers used different cell lines, different peptide sources, and also different solvent systems for administration. One laboratory tested 7 peptide batches from the same and different sources.44 They found high levels of toxicity in only 1 of 7 batches, but could not detect differences in purity or identity based on HPLC, mass spectrometry, or amino acid analysis. This indicates that small, undetectable impurities have a major impact on the process which causes toxicity. Confirmation of Yankner's results came from Cotman and Glabe,45 who showed that there is a correlation between the aggregation state of theP
protein and its neurotoxicity in cell culture. This aggregation-dependent toxicity could be induced by aging peptide samples for 2-4 days at >10 jM peptide concentration. These results suggest that aggregatedf
protein may cause cell death in the AD brain.Many other mechanisms have been proposed for cell death in AD, including disruption of ion homeostasis, disruption of the cytoskeleton (tau), and changes in membrane phospholipids, yet none have satisfied all of the observations in AD.46 It has been recognized that the cell death in the AD brain has the classic markers of a chronic inflammatory process.7 These include activated microglia (macrophages), astrocytes expressing interleukin-1, and the presence of a1-antichymotrypsin.7 Also present in close association with
amyloid plaques are the complement proteins C1q and C4d, which are associated with the immune system. Rogers, et al.,46 have shown that the
P
protein directly activates the classic complement cellular toxicity pathway without mediation by immunoglobulins. This pathway, through a series of binding and activation reactions, results in the formation of the multisubunit membrane attack complex, which lyses cell membranes in the immediate vicinity of the antigenic protein. While it is not clear whether aggregation is required for this activation, in vitro experiments were performed with 500-1500 pM stock solutions, suggesting that this peptide may have been aggregated. In summary, theP
protein appears to be neurotoxic in an aggregation-dependent manner, and this neurotoxicity is mediated by the classic complement immune response pathway.The preceding pages have described a general theory for the molecular pathology of Alzheimer's disease, beginning with the most basic observations about the symptoms of the disease which were recorded 90 years ago, and proceeding to a case for the direct involvement of
P
protein amyloid fibrils in the disease process. Most of the understanding of the biology of plaque formation has been unraveled in the last five years, during the course of the work described in this thesis. Our understanding and interpretation of our own work has evolved as we learned more about AD and protein assembly from otherresearchers. This thesis will present our initial work on fibril formation by a peptide derived from a bacterial protein; this demonstrated that amyloid fibril formation was a nucleation-dependent process. It will then describe a few examples of nucleation-dependent protein assembly processes, upon which a
general model for fibril formation can be based. Kinetic experiments with peptide models derived from the C-terminus of the
P
protein provide some additional insight into possible modes of in vivo amyloid formation. Finally, structural studies have provided preliminary evidence upon which a speculatory mechanism for amyloid formation can be based.References
(1) Coni, N.; Davison, W.; Webster, S. Ageing: The Facts; 1984, Oxford University Press, London.
(2) Katzman, R. New Engl.
J.
Med. 1986, 314, 964.(3) Alzheimer, A. Allg. Z. Psychiatr. Psych. Gerichtl. Med. 1907, 64, 146. (4) Jarvik, L.; Greenson, H. Alzheimer. Dis. Assoc. Disord. 1987, 1, 7.
(5) Rubin, E. in Alzheimer's Disease; Wurtman, R.; Growdon,
J.; Corkin, S.
Ritter-Walker, E.; RavenPress, New York, 1990; p 53.(6) Wisniewski, H. in Alzheimer's Disease and Related Disorders; Iqbal, K.; Wisniewski, H. Winblad, B.; Alan R. Liss, Inc., New York, 1988. (7) Selkoe, D. Neuron 1991, 6, 487.
(8) Selkoe, D. Scientific American 1991, 68.
(9) Blessed, G.; Tomlinson, E.; Roth, M. Br.
J.
Psych. 1968, 114, 797.(10) Roher, A.; Wolfe, D.; Palutke, M.; KuKuruga, D. Proc. Natl. Acad. Sci. USA 1986, 83, 2662.
(11) Kirschner, D.; Abraham, C.; Selkoe, D. Proc. Natl. Acad. Sci. USA 1986, 83, 503.
(12) Birchall,
J.; Chappell,
J.
Clin. Chem. 1988, 34, 265.(13) Candy, J.; Oakley, A.; Klinowski, J.; Carpenter, T.; Perry, R.; Atack, J.; Perry, E.; Blessed, G.; Fairbairn, A.; Edwardson,
J.
Lancet 1986, 1, 354.(14) Miller, D.; Papayannopoulos, I.; Styles,
J.; Bobin, S.; Lin, Y.; Biemann, K.;
Iqbal, K. Arch. Biochem. Biophys. 1993, 301, 41.(15) Abraham, C.; Selkoe, D.; Potter, H. Cell 1988, 52, 487.
(16) Coria, F.; Castano, E.; Prelli, F.; Larrondo-Lillo, M.; van Duinen, S.; Shelanski, M.; Frangione, B. Lab. Invest. 1988, 58, 454.
(17) Yankner, B.; Mesulam, M.-M. New Engl.
J.
Med. 1991, 325, 1849.(18) Glenner, G.; Wong, C. Biochem. Biophys. Res. Comm. 1984, 120, 885. (19) Masters, C.; Simms, G.; Weinman, N.; Multhaup, G.; McDonald, B.;
Beyreuther, K. Proc. Natl. Acad. Sci. USA 1985, 82, 4245.
(20) Joachim, C.; Duffy, L.; Morris, J.; Selkoe, D. Brain Res. 1988, 474, 100. (21) M6ler-Hill, B.; Beyreuther, K. Annu. Rev. Biochem. 1989, 58, 287.
(22) Mori, H.; Takio, K.; Ogawara, M.; Selkoe, D.
J.
Biol. Chem. 1992, 267, 17082.(23) Sinha, S.; Lieberburg, I. Neurodegeneration 1992, 1, 169.
(24) Robakis, N.; Ramakrishna, N.; Wolfe, G.; Wisniewski, H. Proc. Natl. Acad.
Sci. USA 1987, 84, 4190.
(25) Kang,
J.; Lemaire, H.; Unterbeck, A.; Salbaum, J.; Masters, C.; Grzeschik,
K.; Multhaup, G.; Beyreuther, K.; Muller-Hill, B. Nature 1987, 325, 733. (26) Tanzi, R.; Gusella, J.; Watkins, P.; Bruns, G.; St George-Hyslop, P.; vanKeuren, M.; Patterson, D.; Pagan, S.; Kurnit, D.; Neve, R. Science 1987, 235,
8508.
(27) Goldgaber, D.; Lerman, M.; McBride, 0.; Saffiotti, U.; Gadjusek, D. Science
1987, 235, 877.
(28) St George-Hyslop, P.; Tanzi, R.; Polinsky, R.; Haines,
J.; Nee, L.; Watkins,
P.; Myers, R.; Feldman, R.; Pollen, D.; Drachman, D.; Growdon,J.; Bruni,
A.; Foncin, J.-F.; Salmon, D.; Frommelt, P.; Amaducci, L.; Sorbi, S.;Piacentini, S.; Stewart, G.; Hobbs, W.; Conneally, P.; Gusella,
J.
Science1987, 235, 885.
(29) Levy, E.; Carman, M.; Fernadez-Madrid, I.; Power, M.; Lieberberg, I.; van Duinen, S.; Bots, G.; Luyendijk, W.; Frangione, B. Science 1990, 248, 1124. (30) Gandy, S.; Greengard, P. Trends Pharmacol. Sci. 1992, 13, 108.
(31) Schubert, D.; Schroeder, R.; LaCorbiere, M.; Saitoh, T.; Cole, G. Science 1988, 241, 223.
(32) Sisodia, S. Proc. Natl. Acad. Sci. USA 1992, 89, 6075.
(33) Golde, T.; Estus, S.; Younkin, L.; Selkoe, D.; Younkin, S. Science 1992, 255, 728.
(34) Estus, S.; Golde, T.; Kunishita, T.; Blades, D.; Lowery, D.; Eisen, M.; Usiak, M.; Qu, X.; Tabira, T.; Greenberg, B.; Younkin, S. Science 1992, 255, 726. (35) Haass, C.; Koo, E.; Mellon, A.; Hung, A.; Selkoe, D. Nature 1992, 357, 500. (36) Seubert, P.; Oltersdorf, T.; Lee, M.; Barbour, R.; Blomquist, C.; Davis, D.;
Bryant, K.; Galasko, D.; Thal, L.; Lieberburg, I.; Schenk, D. Society for
Neuroscience Abstracts 1992, 18, 165.
(37) Haass, C.; Sclossmacher, M.; Hung, A.; Vigo-Pelfrey, C.; Mellon, A.; Ostaszewski, B.; Lieberburg, I.; Koo, E.; Schenk, D.; Teplow, D.; Selkoe, D.
Nature 1992, 359, 322.
(38) Shoji, M.; Golde, T.; Ghiso,
J.;
Cheung, T.; Estus, S.; Shaffer, L.; Cai, X.-D.; McKay, D.; Tintner, R.; Frangione, B.; Younkin, S. Science 1992, 258, 126. (39) Robakis, N.; Vassilacopoulou, D.; Efthimiopoulos, S.; Sambamurti, K.;Shioi, J. In Proceedings of the The International Study Group on the
Pharmacology of Memory Disorders Associated With Aging; Nitsch, R.;
Growdon,
J.;
Corkin, S.; Wurtman, R.; Ed.; 1993, pp 161.(40) Seubert, P.; Vigo-Pelfrey, C.; Esch, F.; Lee, M.; Dovey, H.; Davis, D.; Sinha, S.; Schlossmacher, M.; Whaley,
J.;
Swindlehurst, C.; McCormack, R.; Wolfert, R.; Selkoe, D.; Lieberburg, I.; Schlenk, D. Nature 1992, 359, 325. (41) Katzman, R.; Saitoh, T. FASEBJ.
1991, 5, 278.(42) Yankner, B.; Duffy, L.; Kirschner, D. Science 1990, 250, 279. (43) Kosik, K.; Coleman, P. Neurobiology of Aging 1992, 13, 535.
(44) May, P.; Gitter, B.; Waters, D.; Simmons, L.; Becker, G.; Small,
J.;
Robison,P. Neurobiology of Aging 1992, 13, 605.
(45) Pike, C.; Walencewicz, A.; Glabe, C.; Cotman, C. Brain Res. 1991, 563, 311. (46) Rogers,
J.;
Cooper, N.; Webster, S.; Schultz,J.;
McGeer, P.; Styren, S.; Civin,W.; Brachova, L.; Bradt, B.; Ward, P.; Lieberburg, I. Proc. Natl. Acad. Sci. USA 1992, 89, 10016.
Chapter 2
Amyloid Fibril Formation is a Nucleation-Dependent Assembly
Process
The presence of amyloid plaques is not unique to Alzheimer's disease. A number of diseases are characterized by the presence of plaques in either a single organ or throughout the body.1'2 Very little is known about the detailed
mechanism of amyloid formation in any of these diseases. Amyloid plaques are usually obtained upon autopsy and, due to their insolubility, provide very little structural information, and virtually no information regarding the mechanism of formation.3 This chapter describes efforts to unravel the mechanism of amyloid fibril formation by hydrophobic peptides. For the purpose of this initial study, we chose a peptide derived from a bacterial protein, the OsmB gene product from E. coli.4,5 This peptide resembles the C-terminus of the
P
protein and is also similar to silk sequences, which also form fibrillar structures.6 After briefly discussing the current understanding of amyloidosis, this chapter will present evidence that amyloid fibril formation is a nucleation-dependent polymerization process. As a nucleation-dependent process, it can be seeded by addition of fibril fragments. In addition, fibril formation is a chemically-specific process, requiring sequence-specific interactions between the peptide and the seed fibril. This type of behavior will compared to two well-characterized protein assembly processes: tubulin polymerization and sickle-cell hemoglobin polymerization. Finally, based upon these examples, the general features, of a nucleation-dependentthe model will be discussed, and the predicted rate curves will be compared to the experimental results.
Amyloidosis
Amyloid Disease
Amyloid diseases are characterized by the presence of insoluble protein deposits in the affected organs. There are more than thirty amyloid diseases, representing a broad range of conditions, both pathologically and biochemically.1 The common feature shared by all of these diseases is the presence of amyloid plaques surrounded by damaged tissue. Rudolph Virchow, upon noting that the plaques stained with a sulfuric acid/iodine protocol, concluded that the plaques must consist of cellulose, thus he coined the term amyloid.7 It was soon proven that the plaques were proteinaceous, but 130 years later, the term amyloid is still in use. The plaques are highly insoluble and resistant to denaturants, and thus are not amenable to biochemical analysis. In the absence of data to the contrary, it was often assumed that the amyloid found in different diseases was the same protein, thus early papers often refer to human amyloid filaments, without citing the protein, disease, or organ source of the preparations.8 In 1970, Glenner and coworkers successfully purified9 and sequenced10 the amyloid protein from a fatal systemic amyloidosis. This protein proved to be a fragment of the immunoglobulin light chain. Since then, fifteen amyloid proteins have been identified, summarized in Table 2.1.2 Based upon the work of Glenner and others, three criteria have come into common usage as definition of amyloid: insoluble proteinaceous material that (1) stains with the dye Congo red, (2) has a fibrillar morphology as viewed by electron microscopy, with 10 nm wide,
Table 2.1
Proteins involved in human amyloid diseases and the apparent cause of amyloid formation. P: proteolysis; M: mutation; C: conformational change. Adapted
from ref. 2, 12.
Distribution Precursor Cause Amyloid Protein Systemic Immunoglobulin (23 kD) P Immunoglobulin (5-23 kD) Systemic Apolipoprotein-SAA (12 kD) P Apo-SAA (8 kD)
Systemic Apolipoprotein-AI (26 kD) M, P Apo-AI (9-11 kD) Systemic Transthyretin (14 kD) M, P, C ATTR (5-14 kD) Pancreas Pro-IAPP (9 kD) P IAPP (4kD) Thyroid Calcitonin (14 kD) P Calcitonin (6 kD)
Muscular P-2-microglobulin (12 kD) C P-2-microglobulin (12 kD) Brain PAPP (110-135 kD) P, M
P
protein (4 kD)Brain Cystatin C (13 kD) M Cystatin C (12 kD) Brain PrP cellular (30-35 kD) C Prion (30-35 kD)
unbranched, twisted fibrils, and (3) shows a characteristic cross-P X-ray fiber diffraction pattern. These criteria, especially the first two, are essential for distinguishing amyloid from other fibrillar material in tissue sections. However, these criteria may not be an indication of any common structure,1 1 this is discussed in more detail below.
In each amyloid disease, the precursor protein has proven to be a soluble or membrane-bound protein which undergoes an alteration resulting in an insoluble variant. In many cases, this alteration is a familial genetic mutation which significantly changes the processing or solubility properties of the precursor, resulting in plaque formation and eventual amyloidosis. For example, in hereditary cerebral hemorrhage with amyloidosis-Dutch type, a substitution of glutamine for glutamic acid at residue 22 of the
P
protein results in formation of vascular amyloid. With short peptides derived from this sequence, the mutated form has been shown to aggregate more rapidly.13 In hereditary cystatin C amyloid angiopathy, the protease inhibitor cystatin C is mutated resulting in a glutamine 68 to leucine substitution. This mutation is believed to result in impaired secretion and intracellular accumulation of cystatin C.2 Amyloidosis in this case may be analogous to inclusion body formation (discussed below). Proteolysis can liberate a small fragment of an otherwise soluble or membrane-bound protein; the fragment may be much less soluble than its precursor, particularly if it is derived from a hydrophobic sequence within the protein. This may be responsible for amyloid formation in AD as was discussed in chapter 1. Finally, a conformational change can result in aggregation of an otherwise soluble protein. The protein transthyretin has been found in both native and proteolyzed states in plaques. Colon and Kelly have shown that the full length protein is soluble; however, the protein forms amyloid fibrils when it is partially denatured at acidic pH, similar to conditions encountered within the lysozome.14A similar type of aggregation-induced conformational change may be responsible for the conversion of cellular prion protein (PrP) to infectious amyloid PrP in the prion diseases.15,16
Mechanism of In Vivo Amyloid Formation.
Very little is known about the details of in vivo amyloidogenesis. Since amyloidosis is a human disease occurring in sensitive organs which are not amenable to biopsy, amyloid plaques are usually only observed upon autopsy. Glenner has studied the structure of systemic amyloid derived from the immunoglobulin light chain. Using X-ray diffraction he studied non-oriented and mechanically oriented amyloid fibril preparations, and observed bands corresponding to 4.75 A and 9.8 A.8 These correspond to the interstrand and intersheet distances in an antiparallel f-sheet. After orienting the fibrils, he showed that the P strands are perpendicular to the fibril axis. On the basis of this evidence, he suggested that the structure is similar to the cross- structure described by Pauling and Corey17 and observed in silk from Chrysopa.18 However, X-ray diffraction does not require that the entire protein be in the cross-n structure, since random, non-repeating structures will not contribute to the diffraction pattern. Glenner also studied these plaques using infrared spectroscopy of the amide I band.19 He studied dried plaques in KBr pellets and as films from 50% formic acid solution. The spectra showed vibrational bands at ~1630 cm-1 which were 40-60% of the total absorption intensity. These are typically assigned as antiparallel
p-sheet.
20 He obtained similar spectra when proteolytic fragments of immunoglobulin proteins were dissolved in 50% formic acid and dried as films. This observation suggests that amyloid formation from immunoglobulin light chain may represent denaturation and aggregation atacidic (lysosomal) pH of otherwise soluble protein fragments. In this respect, amyloid formation may have much in common with in vitro protein aggregation.
The structural and possibly mechanistic link between amyloid formation and in vitro protein aggregation has been reviewed by Wetzel.21 He suggests that the aggregation of proteins during in vitro refolding experiments, the formation of bacterial inclusion bodies, and the formation of mammalian amyloid deposits may all follow similar mechanistic pathways. The current understanding of protein refolding is that a protein first collapses to a molten globule state, possessing secondary structural elements, but little tertiary structure. The molten globule exposes more hydrophobic surface area than the folded protein. If this exposed hydrophobic surface is ordered, then proteins may begin to aggregate in an ordered array, resulting in a stable aggregate. Often, this type of aggregation requires partial denaturation, which exposes some hydrophobic surfaces but allows the protein to maintain a defined structure. For example, tryptophanase aggregates into a highly insoluble aggregate at 3 M urea, but when an 8 M urea solution (completely denatured and soluble) is rapidly diluted into buffer, refolding procedes with very little aggregation.22 This type of denaturation-dependent aggregation was observed for at least two amyloid proteins. Glenner noted that the FTIR spectrum of immunoglobulin amyloid could be reproduced by dissolving the folded immunoglobulin protein in formic acid and then slowly dialyzing into native buffer.19 Colon and Kelly observed that transthyretin aggregated upon partial denaturation in acidic conditions, resulting in amyloid fibrils.14 Some forms of amyloidosis may be similar to inclusion body formation in bacteria. In most cases, inclusion body formation occurs when a protein is overexpressed at concentrations which are higher than its solubility.21 This may be similar to cystatin C amyloid, in which a mutation causes decreased secretion and therefore, intracellular sequestration of the
30
protein above its solubility.2 Studies of protein aggregation have led to several conclusions with consequences for amyloid formation. First, point mutations can cause dramatic changes in aggregation properties by altering the balance of partially folded intermediates.23 Second, certain proteins act in vivo to prevent protein aggregation, particularly the chaperonins (hsp60 and hsp70). These bind weakly to partially folded proteins, preventing aggregation until either secretion or complete folding is accomplished.21 Finally, protein aggregation can be reduced by addition of small molecule inhibitors. These may act in the same manner as chaperonins.24
In summary, amyloid formation generally results from the localized concentration of an insoluble protein or protein fragment. This can be the result of a mutation which results in increased production, altered processing, sequestration, or decreased solubility of the amyloid protein. In other cases, abnormal proteolytic processing may liberate a fragment which has different solubility properties than the original protein. In other cases, amyloid may result from partial denaturation of the amyloid protein in the absence of chaperonins, resulting in aggregation of folding intermediates. In any case, amyloid plaques share properties and structural features with in vitro protein aggregates, suggesting that the in vitro study of amyloid protein folding and aggregation may lead to a better understanding of the disease process.
Fibril Formation by Peptides Derived from the E. coli OsmB Protein
Sequence Similarities Exist in Amyloid Proteins
Amyloid diseases are caused by a wide range of proteins. These differ in sequence and in origin.1 We wondered whether some of these proteins and
possibly other unidentified proteins might share a sequence homology that contributes to their insolubility. Experiments in our laboratory had shown that while many peptides derived from the
P
protein of AD could form fibrils, only peptides derived from the C-terminus shared similar solubility properties with full lengthP
protein.25 In addition, Kurt Halverson showed that a peptide derived from residues 34-42 of theP
protein had an unusual structure centered around a cis-amide between Gly-37 and Gly-38.12,26 He noted that residues 25-40 contained an unusual sequence periodicity: glycine was repeated at every fourth residue. (Figure 2.2) The presence of abnormally high levels of glycine and serine was also noted in immunoglobulin derived systemic amyloid.27 This is reminiscent of silk protein from Bombyx mon, which contains the consensus sequence (GAGS)n6 and of the egg stalk protein of Chrysopa flava, which contain the repeat (GSAS)n.28 It has been proposed that glycine plays a unique role in cross-0 fibrils, allowing the protein to form a bend and zig-zag back and forth across the fibril.29 It is not clear why glycine would be required for this role; its flexibility may be important in maintaining a tight turn and optimal packing. The conformational flexibility of glycine may allow the peptide chain to adopt conformations not normally populated in globular proteins.30 In particular,p-sheets in globular proteins are normally twisted by as much as 15-30' per strand,3 1 while the twist periodicity observed in fibrils suggests that
P
sheets in fibrils twist less than 1' per strand (this assumes that fibrils are really composed of extendedP
sheets, this may not always be the case).12 Twisting ofP
sheets is normally due to the steric interaction of i and i+2 side-chains,31 and the presence of glycine may allow sheets to compensate and remain flat.30Figure 2.2
Sequence of the C-terminus of the
P
protein showing the GXXX repeat, along with the scrapie prion protein and two regions from the OsmB protein.P
protein (29-42):
. G..AIIGLMVGGVVIA-CO2HPrion Protein (96-111):
... AGAVVGGLGGYMLGSA...
OsmB Protein (14-25):
..
.GAGAGALGGAVL...
OsmB Protein (28-44): ... GSTLGTLGGAAVGVIG...
We developed a set of criteria in order to search for proteins homologous to the C-terminus of the
P
protein. (see experimental section for details) We insisted on at least 3 contiguous repeats of the GXXX repeat observed in residues 29-40 of theP
protein. We reasoned that the X-residues would be located in the core of the fibril and would therefore be uncharged and would not be proline, which disruptsP
structure. Thus, we limited X to G, A, V, I, L, F, W, Y, T, S, or M. Eighty sequences were obtained upon searching the sequence database. Sequences containing polyglycine and long GXGX repeats (silk proteins) were eliminated. We also reasoned that residues which are common in the -sheets found in globular proteins would be likely to populate the 1-sheet structures which are characteristic of many amyloids.30 The remaining sequences were screened for those that were relatively hydrophobic (hydropathy 2 1.4),32 and contained residues often found in 1-sheets but rarely in c-helices in globular proteins(Pp-Pa 0.17).33 Twenty-seven sequences survived including theP
amyloid protein (res. 29-40), the scrapie prion protein (res. 96-111), and the E. coli OsmB protein (res. 14-25 and res. 28-44) (Figure 2.2).The E. coli OsmB protein contains two similar sequence repeats, comprising ca. 70% of the 49 residue protein.4 OsmB is a periplasmic outer membrane-associated lipoprotein, which is upregulated in response to osmotic
Figure 2.3
Sequence of the peptide derived from OsmB(28-44) and two controls, OsmG3 and OsmA.
OsmB(28-44): AcNH-GSTLGTLGGAAVGGVIG-CONH2
OsmG3: AcNH-GSTGLTGLAGAVGVIGG-CONH2
OsmA: AcNH-GSTLATLGAAAVAGVIG-CONH2
stress. Its function is unknown, but it has been suggested that it plays a role in reinforcing the outer membrane, preventing cell death during sudden osmotic changes.4 We synthesized a peptide corresponding to residues 28 through 44 of the OsmB protein (Figure 2.2 & 2.3) in order to investigate the hypothesis that this sequence motif drives amyloid formation. Our preliminary findings were that the peptide (OsmB(28-44)) was initially soluble in water (~1 mg/ml), but formed a solid when the solution was stirred or allowed to stand for several days. This solid showed fibrillar morphology by electron microscopy. (Figure 2.3A) Since our initial search was based on the hypothesis that the glycine repeat was important in fibril formation, we synthesized two control peptides to test this hypothesis. OsmG3 had the same amino acid content as OsmB(28-44), but the sequence was minimally rearranged to generate a sequence which had glycine at every third residue. (Figure 2.3) OsmA was a "mutated" form of OsmB(28-44) in which three of the glycines were replaced with alanine.
OsmB Derived Peptides Form Amyloid Fibrils
Metastable supersaturated solutions of each peptide could be prepared in water. These solutions were stable for hours to days, depending upon the degree of supersaturation. Agitation of these solutions by stirring or sonication caused
aggregation to occur with little or no delay. The degree of aggregation could be followed by measuring the turbidity of the solutions, and generally approached an equilibrium within 1-2 hr. The aggregates were studied by electron microscopy and Fourier-transform infrared spectroscopy, and tested for Congo red staining and birefringence.
As viewed by electron microscopy, (Figure 2.4) the morphology of the peptide fibrils varied slightly. In general, amyloid fibrils are expected to form long unbranched fibrils, 50-100
A
in diameter, with a periodic twist, and should associate in parallel bundles. OsmB(28-44) formed twisted fibrils 100-150 A in diameter, with a twist periodicity of 1300 ± 100 A. These fibrils aggregated in two distinct forms, depending on peptide concentration. In more dilute samples (<500 pM), two fibrils twisted around each other in a helical manner with a periodicity of 1800 A. In concentrated samples, the fibrils tended to clump into ribbon-like structures containing 20-30 fibrils, with each ribbon about 200 x 1000A
in cross section and several thousand nanometers long; these ribbons also clumped strongly together. OsmG3 formed fibrils which were similar to those from OsmB(28-44) with respect to width and twist periodicity, however the OsmG3 fibrils tended to wind together in larger fibrils containing between 2-5 smaller fibrils. OsmA formed straight rods, also about 100 A in diameter, with no observable twist. These clumped together in a parallel manner, but with no distinguishable higher-level order.The aggregates of each peptide were also tested for Congo red binding and birefringence. All of the peptide aggregates bound Congo red, resulting in reddish-pink films. After staining, both OsmB(28-44) and OsmA exhibited yellow-green birefringence, however, stained OsmG3 was not birefringent. As was mentioned above, Congo red binding and birefringence, along with morphological examination of the fibrils, is used to histochemically define
Figure 2.4
Electron micrographs of (OB) OsmB(28-44) left formed at 425 PM, right formed at 1 mM; (OG) OsmG3; (OA) OsmA, both formed at 425 pM. Bar = 1000 A.
OB
.~ ~ . $
.....
